Discussion
Costly sexually selected traits may affect viability, especially in stressful environments, and it has long been speculated that, for example, Irish elk (Megaloceros giganteus ), characterized by extremely elaborated antlers, went extinct when glaciation limited resource availability (Moen et al., 1999; but see O’Driscoll Worman & Kimbrell, 2008). However, the effect of elaborated sexual traits on extinction risk remains unresolved, with comparative analyses yielding conflicting results (Doherty et al., 2003; Morrow & Pitcher, 2003; Martins et al., 2018; Parrett et al., 2019). Our study is the first experimental evidence showing that during the environmental change, the presence of elaborated sexual trait in a population elevates the risk of extinction.
Our results contrast with those obtained by manipulating opportunity sexual selection via mating system and sex ratio, which generally find positive effect of sexual selection on fitness correlates (reviewed in Cally et al., 2019) and population survival (Jarzebowska & Radwan, 2010; Plesnar-Bielak et al., 2012; but see Parrett and Knell 2018). This apparent inconsistency can be reconciled by conceptually separating short-term from long term impacts of sexual selection (Figure 3). In the short timescale, the effects of sexual and natural selection act on phenotypic variation present in a population at a given point in time. Under condition-dependence, both types of selection can to a large extent be aligned, especially during environmental change (Long et al., 2012; Plesnar-Bielak et al., 2012; Parrett & Knell, 2018). However, in the long term, as sexually selected traits get elaborated, extinction risk may be increased, for example by magnifying survival costs to males (Promislow, 1992; Moen et al., 1999), or by causing gender load associated with (partial) expression in females of alleles favoured by sexual selection acting on males (Rice & Chippindale, 2001; Berger et al., 2016).
Our results do not support the hypothesis that rate of extinction of entire populations is increased because cost of developing and carrying heavy weapons by males increase their mortality under environmental challenge (Moen et al., 1999; Kokko & Brooks, 2003). Increasing temperature did not disproportionately affect survival of males in FT populations. Instead, in both FT and ST populations, females survived better than males in early generations, but in generations 3 and 4 the differences between the sexes disappeared, suggesting that female survival was more negatively affected by increasing temperature compared to males (Fig. S3). Our results support earlier suggestions that that because of their widespread condition-dependence, elaborate sexually selected traits should not compromise population fitness because most males not express these costly traits when environment deteriorates (Kokko & Brooks, 2003). This was indeed the case in the present study, as proportion of fighters decreased across generations with increasing temperature in FT populations, with proportion of fighters already down to 75,7% at the first step of temperature increase (Figure S2). This decrease can be explained by condition-dependence weapon expression (Radwan, 1995; Smallegange, 2011; Plesnar-Bielak et al., 2018), and, additionally by selection against ‘fighter genes’ under increased temperature (Plesnar-Bielak et al., 2013). Irrespective of the cause, suppression of weapon expression with increasing thermal stress did not prevent extinctions. Indeed, most of extinctions occurred at generation four, when only a minority (25%) of males expressed the costly weapon. Thus, our results demonstrate that evolution of costly, sexually selected traits may affect the risk of extinction even when the expression of such traits is reduced due to their condition-dependence.
Thus overall, survival cost of developing and maintaining elaborated sexual traits by males does not appear to be a reason for increased extinction of FT populations. Another reason for elevated extinction risk of fighter populations could be differences in effective population sizes (Ne), which might occur if reproductive success among fighter males is more biased compared to scrambler males. Parrett et al. (in prep) have estimated Ne for populations fixed for scrambler and fighter morph at 37% and 46% of the census population size, based on SNPs frequency changes between generations. This implies that for our populations of 50 individuals, Ne would be 18.5 and 23.0, respectively, and the resulting increase in inbreeding over four generation, assuming first generation was outbred, 0.08 and 0.10. A difference in inbreeding of the order of 2% is unlikely to explain a significant proportion of the difference in fitness and extinction risk between our treatments. Yet another possible reason is increased sexual antagonism associated with male weapon, which in the bulb mite was reported to be genetically correlated with decreased female fecundity and survival (Plesnar-Bielak et al., 2014; Łukasiewicz et al., 2020). Therefore, sexually antagonistic, pleiotropic effects of male weapon on female fitness may have contributed to increased extinction risk under genetic or environmental stress. The role of ontogenetic conflict in extinction was implied in the study of bean beetles, where inbred lines associated with high male fitness, but low female fitness, suffered increased extinction risk under inbreeding (Grieshop et al., 2017). However, that study did not eliminate the possibility that high-male-fitness inbred lines carried higher load of deleterious recessives. In case of our study, such explanation could be ruled out by significantly lower inbreeding depression we recorded in our fighter lines after four generations of inbreeding (Łukasiewicz et al., 2020). The same study confirmed that gender load was present, manifested as decreased fecundity in outbred females descended from F-lines compared to those derived from S-lines. This gender load may have interacted with environmental stress, thus increasing extinction rate, as earlier suggested by Grieshop et al. (2017) for genetic stress. Finally, F populations might have failed to adapt to increasing temperature if deleterious variants removed from F populations via enhanced “good genes” selection (Łukasiewicz et al., 2020) were rendered adaptive under increased temperature (Jensen, 2014). Further work is needed to discriminate between these alternatives.
It has been suggested that the increase in sexually selected dimorphism may increase adaptive potential of populations by helping to maintain genetic variance (Radwan et al., 2015) or by increasing environmental scope (Bonduriansky, 2011; De Lisle & Rowe, 2015). Our results suggest otherwise, but the outcome may be context-dependent and affected by e.g. the rate of environmental deterioration and population size. Large populations can better preserve genetic variation under negative pleiotropy than small populations (Connallon & Clark, 2012), and thus large populations of sexually dimorphic species may be able to use sexually antagonistic variation to respond to environmental changes. Furthermore, selection is more effective in larger populations, and ‘good genes’ effects may prevail over negative pleiotropic effects (Martinez-Ruiz & Knell, 2017). Further work is required to elucidate the role of costly, sexually selected trait on extinction at various demographic and environmental scenarios.
Acknowledgements: We thank Joe Tomkins and Jon Parrett for their comments on earlier versions of this manuscript, and Katerina Altouva for help during experiments. This work was supported by NCN grant UMO-2020/39/B/NZ8/00152to JR